Hydrological performance of extensive green roofs in New York City: observations and multi-year modeling of three full-scale systems

Abstract

Green roofs can be an attractive strategy for adding perviousness in dense urban environments where rooftops are a high fraction of the impervious land area. As a result, green roofs are being increasingly implemented as part of urban stormwater management plans in cities around the world. In this study, three full-scale green roofs in New York City (NYC) were monitored, representing the three extensive green roof types most commonly constructed:(1) a vegetated mat system installed on a Columbia University residential building, referred to as W118; (2) a built-in-place system installed on the United States Postal Service (USPS) Morgan general mail facility; and (3) a modular tray system installed on the ConEdison (ConEd) Learning Center. Continuous rainfall and runoff data were collected from each green roof between June 2011 and June 2012, resulting in 243 storm events suitable for analysis ranging from 0.25 to 180 mm in depth. Over the monitoring period the W118, USPS, and ConEd roofs retained 36%, 47%, and 61% of the total rainfall respectively. Rainfall attenuation of individual storm events ranged from 3 to 100% for W118, 9 to 100% for USPS, and 20 to 100% for ConEd, where, generally, as total rainfall increased the per cent of rainfall attenuation decreased. Seasonal retention behavior also displayed event size dependence. For events of 10-40 mm rainfall depth, median retention was highest in the summer and lowest in the winter, whereas median retention for events of 0-10 mm and 40+ mm rainfall depth did not conform to this expectation. Given the significant influence of event size on attenuation, the total per cent retention during a given monitoring period might not be indicative of annual rooftop retention if the distribution of observed event sizes varies from characteristic annual rainfall. To account for this, the 12 months of monitoring data were used to develop a characteristic runoff equation (CRE), relating runoff depth and event size, for each green roof. When applied to Central Park, NYC precipitation records from 1971 to 2010, the CRE models estimated total rainfall retention over the 40 year period to be 45%, 53%, and 58% for the W118, USPS, and ConEd green roofs respectively. Differences between the observed and modeled rainfall retention for W118 and USPS were primarily due to an abnormally high frequency of large events, 50 mm of rainfall or more, during the monitoring period compared to historic precipitation patterns. The multi-year retention rates are a more reliable estimate of annual rainfall capture and highlight the importance of long-term evaluations when reporting green roof performance.

Detention and Retention Behavior of Four Extensive Green Roofs in Three Nordic Climate Zones

“…This is confirmed by several studies showing that the fraction of water retained decreases as the event sum precipitation increases [21,26,27] and that retention increases as substrate initial moisture decreases [28]. Significant improvements in detention performance are also found at lower initial substrate moistures [29], mainly due to the effect of initial losses by retention, but possibly also affected by the fact that hydraulic conductivity increases as soil moisture content increases.…”

“…There is a large inter-event variability shown in most reported studies with reported retention from 0 to 100%, showing a decrease in retention with increasing event sum precipitation [9,21,26,27], but also, other factors such as antecedent soil moisture content are found to be important.…”

“…Few actual field studies, and especially long duration studies, are available from cold and wet climates comparable to the Nordic climates, and design estimates for retention are needed. Most studies from locations with large seasonal variation in evapotranspiration report lower retention in winter and higher in summer [27,31,[38][39][40]. Some studies does also explain part of the observed seasonal variations as a result of seasonal variations in precipitation patterns [34,41].…”

“…The median (min/max) or the mean (bold) ± the standard deviation given. The concept of dividing data into events based on minimum antecedent dry periods and comparing different event metrics is widely applied and efficient when handling large time series [21,27,34]. For the evaluation of retention performance, this is useful since it allows for a closed water balance, but only if event duration is short enough not to include any significant AET.…”

A Simplified Model for Modular Green Roof Hydrologic Analyses and Design

“…Volume reduction decreases as precipitation depth increases (Figure 4a). Runoff was stated to follow a second order polynomial equation in several studies [19][20][21]. This relationship works well for describing slow underflow drainage generated by relatively light precipitation, but would not apply for heavy precipitation or with a water storage reservoir installed.…”